Jobs Count Special Edition: Jobs (Mis)Count
Each month, we put out our Jobs Count publication, which reports on the state's jobs numbers, using data from the Bureau of Labor Statistics Current Employment Statistics (CES). Each month, the BLS revises the previous month's numbers, but the revisions are typically small, and usually don't change the overall picture. However, at the beginning of each year, the BLS does a big revision, known as benchmarking. This year state numbers were revised back to 2008, and this time the revisions completely changed West Virginia's employment picture.
Before the revision, and as reported in our last edition of Jobs Count, West Virginia's total non farm employment stood at 746,900 jobs in December. But after the revision, December's jobs count was 765,500 jobs. The revision added 18,900 jobs to West Virginia, an increase of 2.5%. While most states saw an increase in total jobs with the revision, West Virginia had the second largest revision among the fifty states.
In fact, almost all of West Virginia's employment sector's saw significant revisions. The mining and logging sector had the biggest change, with the revision adding 3,100 jobs to December's count, an increase of 10.5%. Only the education and health services sector saw a downward revision, subtracting 100 jobs.
So what do the revision's mean for our previous Jobs Count reports? Most of our analysis for the past year+ is no longer accurate. For example, in our last edition, we reported that the state was 13,400 jobs below its pre-recession level in December, when after the revision, the numbers show that West Virginia was actually 5,500 jobs above its pre-recession level.
As the chart shows, total non farm employment grew faster in 2011 than initially reported, and the big job losses of 2012 were much smaller and lasted a much shorter amount of time.
So when we reported that 2012 was shaping up to be one of the worst years for job growth since the recession, with 10 straight months of job losses, the revisions completely changed that. Instead of the state losing 13,900 jobs in 2012, with major losses for government and mining; the state gained 1,400 jobs, with increases coming from education and health services and leisure and hospitality.
Here's another piece of analysis that looks completely different. On the five year anniversary of the recession, we noted that job growth had been mixed, with the state still below pre-recession levels. Now, after the revisions, we can see that the sectors that have lost jobs since the recession began have lost fewer jobs than we thought, while the sectors that gained jobs have gained more than we thought.
Because of the size of the revision, the narrative of the past two years of Jobs Counts reports has been off. What initially looked like a double dip recession in 2012 was actually slow growth. With the monthly estimates being this inaccurate, the Jobs Count reports lose their usefulness, and we're doing a disservice in continuing to produce them. Going forward, we're likely to change how we do Jobs Count to avoid these problems. On possible solution would be to use the Quarterly Census of Employment and Wages (QCEW) rather than the CES. The QCEW is more accurate and provides greater detail (for example, you are able to separate coal mining from natural gas drilling in the QCEW but not in the CES), but in exchange for more accuracy, the data are several months old, and are released quarterly, not monthly. But if we are interested in what is happening to the state's jobs market, accuracy is more important than timeliness. As John Maynard Keynes (might have) said about changing his position on an economic issue, "When my information changes, I change my mind. What do you do?"